Automation with purrr

class: title-slide, left, bottom

# Automation with purrr
----
## **or how we created over 3000 pdf letters!**
### Josephine Browning
### 11/11/2020

---
class: center,middle

# About Me

## Josephine Browning

### Business Intelligence Manager at Gloucestershire CCG

----
.left[I have over 10 years experience as an NHS Analyst working across Public Health, Acute Providers and Commissioning Organisations]

.left[I am an NHS-R community trainer and an AphA professionally registered analyst]

.left[I am also currently studying for an MSc in Health Data Science at Swansea University]

----
[ @randomrainbow86](https://twitter.com/randomrainbow86) 
[ josephine.browning@nhs.net](mailto:josephine.browning@nhs.net) 
🎉

---
class: inverse, middle, center

# Why write 3000 letters?

---

# Shielded Patients Welfare Calls

----
### In Gloucestershire we had over 10,000 patients that were identified as requiring to shield

--
### To help some of our GP's we set up a call centre to contact all of the identified patients for a welfare check
--

### Patients were asked a series of questions to ensure they understood what shileding was, why they were being asked to do it and if they needed any help to allow them to shield
--

### Following each call a transcript of the conversation was required to be sent to the GP practice to be added into the patients primary care record
--

### That was where we came in... how do we write all of those PDF's without having to compile each one separately?
---

# Using R Markdown to create a template
----

### The first task was to create a letter template that would be populated for each patient

### Using R Markdown this was written to output to a pdf file format

### Utilised the inline code function to input the answers to each question

### Utilised the Params functionality to input each patients details

---
#Inserting an SQL - R link

----
###To help with the automation the data was pulled into R directly from the SQL data warehouse

--
###This was done individually within the R Markdown template for each patient (This could probably have been coded better to bring all the data in at the start)

--
###We also used this link code to create a dataframe to run the Purrr function over

---
#Using Purrr

----
###The [Purrr](purrr.tidyverse.org) package is part of the tidyverse and has a quite a few useful functions

--
###Map can be used as a nice tidy way to code for loops

--
###For this task we used the Walk2 function as we did not need to see the outputs as it was running

---
#Putting it all together

.panelset[
.panel[.panel-name[Bullets]
###The R-Markdown Template needed an NHS number to produce the letter for that patient

###The Walk2 function needed a dataframe of NHS numbers and GP practice codes and also a function to iterate over

###The output was that a PDF letter was created for each NHS number and this was saved down into a separate folder for each GP practice
]
.panel[.panel-name[Render Report]

```r
render_report <- function(NHS_Number, Practice) {

filename <- paste0("Sample/folder/",Practice,"/",NHS_Number,".pdf")
 
 render(input = "Script1.Rmd",
 output_format = "pdf_document",
 output_file = filename,
 params = list(NHS_Number = NHS_Number))
}
```
###The above code is used to run/render the R Markdown letter template for a given NHS Number]
.panel[.panel-name[Walk2]

```r
library(rmarkdown)
library(purrr)
walk2 (.x = data$NHS_Number, .y = data$Practice_Code, .f = render_report)
```
###The above code produced a PDF for the each NHS number in the dataframe saved in a file for each practice code
]
]
---
class: center, middle
#With thanks to:

##Anna Schorfield and Lee Tarbuck